Simulation model for swine lagoon sludge drying in solar-assisted greenhouse systems

Greenhouse drying is gaining acceptance in agricultural waste applications due to low cost, low energy inputs, and simple operation. This study monitored sludge drying in two mechanically ventilated greenhouses in Eastern North Carolina, USA (488.9 m 2 each) with ventilation automatically controlled in one greenhouse (ACGH), and manually controlled in the other (MCGH). Excavated swine lagoon sludge was loaded into ACGH and MCGH with respective loading rates of 158 and 177 kg m −2 and dried under air exchange rates between 1.1 and 4.6 m 3 m −2 min −1 . The goal was to assess and predict local weather and greenhouse operation impacts on drying. Statistical models (multiple linear regression, decision tree, absolute penalty neural network) were developed to predict drying rates. Results showed two drying phases over the 34-day drying period in both greenhouses. Average drying rates during phases I and II were respectively 4.3 and 0.2 kg H2O m −2 d −1 (ACGH) and 4.5 and 0.1 kg H2O m −2 d −1 (MCGH). In descending order, the drying rate was significantly influenced by solar radiation>sludge moisture>ambient temperature and relative humidity> time since mixing> ventilation rate. The absolute penalty neural network model outperformed other models and was effective in predicting drying rates in both MCGH and ACGH (R 2 > 0.91, RAS (root averaged square error) E < 33.5 kg H2O hr −1 ). This study is first of its kind to evaluate feasibility of greenhouse drying of animal lagoon sludge in Southeastern US. Findings and models developed in this study will guide process optimization and incentivize on-farm greenhouse drying adoption. • Greenhous Drying under manual and automated ventilation was evaluated. • Drying rate was impacted by weather ( T amb , RH amb , RAD) , sludge moisture, mixing. • Average drying rates over a 34-day drying ranged from 2.2 to 2.9 kg m −2 d −1 . • Neural network (NN) model best predicted drying rate under manual ventilation. • NN model successfully predicted automated ventilation drying rates.